Sign up free

Audio and video quiz questions: captions and transcripts

What accessibility guidelines ask of recorded clips, what studies found, how exam bodies handle listening tests, and how to write a transcript that tells no answer.

The EdProcto editorial teamWritten and checked against the research it cites. Last checked . How the guides are made

Guides · Writing questions that measure · · 13 min read

How do I make audio and video quiz questions work for students who cannot hear or see the clip?

Audio and video quiz questions need a second route for a student who cannot hear or see the clip. Where the clip only delivers content, give captions a person has checked, a transcript and a description of what is seen. Where hearing or seeing is itself the skill tested, agree another arrangement beforehand, as examining bodies do.

Audio and video quiz questions ask every student to hear or see a clip before they can answer. A student who is deaf or hard of hearing, or blind or partially sighted, needs what the clip carries in another form: captions, a transcript, a description of what is on screen.

One case is harder. Where the listening or the looking is itself what is tested, the transcript is the answer, and a fair arrangement has to be agreed, not typed.

EdProcto, which publishes this guide, builds clip questions such as listen and answer and watch and answer. A video question there takes a captions file of its own, and the guide says how that fits with a transcript. Every source was read on 10 October 2026; "abstract read" marks a study read only in its abstract. Nothing here is legal advice: entitlement is for your institution to decide.

What do the accessibility guidelines ask for captions, transcripts and audio description?

They ask for a route in text, or in added speech, to everything a recorded clip says and shows. The Web Content Accessibility Guidelines (WCAG) 2.2 come from the World Wide Web Consortium (W3C), which explains each requirement on its own page. The table follows those pages as checked on 10 October 2026; Level A is the lowest of three.

The clip What is asked for Level Explained in
Audio alone A transcript A W3C (2024b)
Video with no sound A descriptive transcript, or a spoken track that describes the picture A W3C (2024b)
Video with sound Captions A W3C (2026b)
Video with sound Audio description, or a full version in text of sound and picture A W3C (2026c)
Video with sound Audio description AA W3C (2026c)
  • Captions carry more than dialogue. They say who is speaking and include sounds that mean something (W3C, 2026b).
  • A transcript is no substitute for captions on a video with sound. By itself it sits at Level AAA, the highest; captions are Level A (W3C, 2024b).
  • A descriptive transcript adds what is seen, and is the form a student who is both deaf and blind needs (W3C, 2024b).
  • Description is not always needed. Where the sound track already gives everything important in the picture, none has to be added (W3C, 2026c): say what your hands are doing as you film.

Where do the guidelines make room for a test?

On the pages we read, in one criterion. The first, on content that is not text, says that where a test or exercise "would be invalid if presented in text", the text alternative need only identify the content. The World Wide Web Consortium (2026a) gives a hearing test, an exercise in visual skill and a spelling test as examples.

The pages on captions and on description each name one exception, a clip that only repeats text already on the page and is labelled so, and neither makes an exception for a test (W3C, 2026b, 2026c). How the two fit together for a listening examination those pages do not settle, and nor does this guide: ask whoever answers for accessibility where you teach.

What does research say about captions and transcripts?

It says captions help hearing viewers too, and that for deaf and hard-of-hearing students text gives access to the words without closing the gap.

Hearing students

Dallas et al. (2016) ran two video-based studies at one university, assigning groups of undergraduates without disabilities at random to captions or none. The captioned groups scored significantly higher on the assessment that followed (abstract read; no figures given). Our video questions guide reports a larger review pointing the same way.

Deaf and hard-of-hearing students

  • Text is not equal access. Marschark et al. (2006) had 95 deaf and 32 hearing university students watch a lecture. The deaf students had a sign language interpreter, a running text typed as the lecture was given, or both, and averaged 53%, 67% and 56% on a 15-question test; the hearing students, with no support, averaged 84%. Three further experiments found no significant difference between kinds of support.
  • Captions are a reading task. In Jelinek Lewis and Jackson (2001), scores on captioned television followed reading level closely, deaf students scored below hearing students in every caption condition, and they understood more when the picture came with the captions (abstract read).
  • Speed matters. Tyler et al. (2009) showed deaf school pupils documentaries captioned at 90, 120 and 180 words a minute: comprehension was higher at the two slower rates (abstract read).
  • Automatic captions need a person. Parton (2016) counted 525 phrase-level errors, 7.7 a minute, in the automatic captions on 68 minutes of lecturers' video updates (abstract read).

So correct captions by hand, keep the picture in view and allow reading time. None of these studies was of a quiz, so they give a direction and no figure.

What do you do when the listening itself is what is tested?

Keep the words off the page for everyone, and agree a different arrangement for the student who cannot hear the clip. Two examining bodies' public pages, checked on 10 October 2026, show how; neither gives a candidate the words of a listening test to read.

Body A candidate who cannot follow the recording Where that is not enough
IELTS (n.d.) A lip-reading version: a supervisor reads each text aloud twice, pausing in the second reading for answers Exemption from Listening, with a score worked out from the other sections and a statement on the report form
Cambridge English (n.d.) The same version, read two or three times, with two readers for dialogues in some exams; or a recording stopped before, during and after each part Exemption, marked on the certificate beside the component

England's primary school tests go further for their spelling paper: a pupil with a profound hearing impairment who does not lip-read or sign may be awarded compensatory marks, based on the average raw score of pupils who took it (Standards and Testing Agency, 2026).

IELTS (n.d.) treats sight the same way: its large-print papers leave out visual material not needed to answer and may give a graph as text in a table, and it offers text or Braille descriptions of photographs.

For a class quiz, sort each clip question first.

The question tests The class gets A student who cannot hear or see the clip gets
Content the clip delivers: a recorded lecture, a demonstration Captions, and a transcript with the question: in EdProcto, the material under the prompt The same, with time to read
Listening itself: a language, a heart sound No text until the quiz is over A live reading to lip-read, another question on the same aim, or an agreed exemption
Seeing itself: a technique, a specimen No description that names it A question in words on the same aim, or an agreed exemption

How do you write a transcript that carries what the question needs?

Write down everything said, and every sound and sight the question rests on, and stop short of the judgement the student is asked to make. The steps follow the W3C's (2020) guidance on transcribing, which calls them common practice, not requirements.

  1. Correct a script or an automatic draft against the clip. In the W3C's own example, four to five minutes of cooking came out as forty-five.
  2. Name each speaker.
  3. Put sounds that matter in brackets: (kettle clicks off).
  4. Keep the speaker's words, grammar slips included, and write [unintelligible] where a word is lost.
  5. For a video, add what is seen and not said: actions, and words on screen. The W3C (2024b) sets these in two columns, audio and visual, and advises describing in the present tense without interpreting (W3C, 2024a).
  6. Answer the question from the transcript alone. It should take the same knowledge as watching did.

A worked example: one clip described two ways

The clip is made up for this guide: twenty seconds of video from a school laboratory.

Audio Visual
(water pouring) A student pours water into a measuring cylinder on the bench.
Student: "Twenty-four." She stays standing and looks down at the scale from above, then writes in a notebook.

The question: why might her reading be wrong? A full answer says her eye was above the water's surface, not level with it, so she read the scale at an angle.

A second version of the last cell: "She reads the scale carelessly, from the wrong height." The first says where her eyes were. The second says it was a mistake, and a student need only repeat it. Describe the action; leave the verdict to the answer.

How do you describe a picture without giving the answer away?

Say what the picture shows, and leave out what the question asks the student to conclude. Such a description is often called alt text. The fullest guidance we found is an archived project page of the National Center for Accessible Media (2026), for test developers and teachers; it says formal research on describing images for tests was only beginning. Its main points:

  • Write the description with the question, not after the test is finished.
  • Describe every picture in a question alike, answer options included, in one style and at about one length, so nothing marks out the key.
  • Be brief. A listener cannot skim, and tires.
  • Give data as data: a chart as a table, a process as a list.
  • Have it read by someone who has not seen the picture.
Picture and question Gives the answer Carries the picture
A bar chart of rainfall. "Which month was wettest?" "Bar chart: July was the wettest month." "Bar chart of rainfall in millimetres. April 40, May 55, June 30, July 80."
A photograph of a dividing cell. "Which stage is this?" "A cell in anaphase." "A cell under a microscope. Two groups of dark threads have moved apart, one towards each end."

Where recognising a thing on sight is the skill, as with a leaf or a road sign, any description changes the question: see picture choice questions.

What do you arrange for a student who needs another form?

Arrange it before the quiz, with whoever agrees such things where you teach.

  1. Ask early. Entitlement is your institution's decision: see extra time.
  2. Sort each clip question with the table above, and prepare its form: corrected captions, a transcript or a description for content; the agreed alternative for a skill. In EdProcto the alternative goes in a copy of the quiz.
  3. Add reading time. Students working from descriptions usually need longer (National Center for Accessible Media, 2026), and Cambridge English (n.d.) lets deaf candidates ask for up to 25% more time in its reading and writing papers.
  4. Rehearse. Let the student try a practice question with their own screen reader or hearing equipment.
  5. Keep a record: the student, the question, what was given and who agreed it.

How does EdProcto handle audio and video quiz questions today?

EdProcto is the best tool for clip questions when some students need another form, because the words are kept with the question: each question has its own clip, its own material box and, for a picture, its own description, and a version for one student is a copy with its own code. The guidance above asks that a description be written with its question; here both are written in one editor.

  • The clip. The ready shapes that take a clip include listen and answer, listen once, dictation and watch and answer, with Pick the clip, whose options are clips, and Compare two clips. The question takes "Audio clip address" or "Video address" and the switch "Plays once, no replay".
  • The words. "Material under the prompt (optional)", opened where it is not already there with "Add material under the prompt", keeps its line breaks and is read with the question: the place for a transcript where the clip only delivers content. For a listening test, leave it empty and paste the transcript into "Explanation shown to the student afterwards".
  • The picture. With "It is" at "An image" and the address filled in, "Describe the picture (for a screen reader, optional)" takes up to 300 characters.
  • Captions. A video question takes "Captions file address (.vtt, optional)", and the captions show on the video, which the student can turn off in the player. Have automatic captions corrected by a person before you link the file. The "Accessibility" settings also offer "Screen-reader mode", "Larger text", "High contrast", "Audio for visual questions" and "Translate into the student’s language".
  • Another form for one student. "Duplicate this quiz", at the foot of the builder, makes a copy with its own code: change the question or add the transcript, raise "Minutes overall", and give that link to the student alone.

Try the quiz yourself with the sound off, and with a screen reader, before the class meets it. Free has ten quizzes a month, each copy counting as one, and thirty students a quiz; Tutor opens all 111 templates for 100 students a quiz, and Educator takes 500. See pricing.

Quick questions

What is the difference between captions and a transcript?

Captions are text timed to a video, with who is speaking and the sounds that matter. A transcript is that text set out to read by itself, and a descriptive transcript adds what is seen (W3C, 2024b).

Does a transcript give away the answer to a listening question?

Only where hearing is what is tested. Then hold it back until afterwards and agree another arrangement: IELTS (n.d.) and Cambridge English (n.d.) use a live reading for lip-reading, or an exemption.

Are automatic captions good enough for a quiz?

Not unchecked. Parton (2016) counted 7.7 phrase-level errors a minute in automatic captions on lecturers' videos, so have a person correct them.

Does EdProcto add captions to a video?

From a file you give it. In EdProcto, a video question takes the address of a captions file (.vtt), and the captions show on the video; a transcript goes in "Material under the prompt (optional)".

References

The settings

Time, attempts, feedback, grading, access, identity, security, tools, accessibility: every rule of a quiz, set by itself.

On every plan

See the settings

Read next